Ensemble Techniques Project

Part A

A telecom company wants to use their historical customer data to predict behaviour to retain customers. You can analyse all relevant customer data and develop focused customer retention programs.

Importing modules for project

import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
from sklearn.preprocessing import LabelEncoder
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from sklearn.neighbors import KNeighborsClassifier
from sklearn.linear_model import LogisticRegression
from sklearn.naive_bayes import GaussianNB
from sklearn.svm import SVC
from sklearn.tree import DecisionTreeClassifier
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import accuracy_score, confusion_matrix
from sklearn.metrics import classification_report
from sklearn import model_selection
from sklearn.metrics import accuracy_score, confusion_matrix
from sklearn.metrics import classification_report
from sklearn.model_selection import RandomizedSearchCV, GridSearchCV
from sklearn.metrics import roc_auc_score
from sklearn.model_selection import StratifiedKFold  
from sklearn.model_selection import cross_val_score  
import pydotplus
import graphviz
from sklearn import model_selection
import warnings
warnings.filterwarnings("ignore")
import pickle
import os
os.environ["PATH"] += os.pathsep + 'D:/Program Files (x86)/Graphviz2.38/bin/'

Read ‘TelcomCustomer-Churn_1.csv’ as a DataFrame and assign it to a variable.

df1=pd.read_csv("TelcomCustomer-Churn_1.csv")

df1

	customerID	gender	SeniorCitizen	Partner	Dependents	tenure	PhoneService	MultipleLines	InternetService	OnlineSecurity
0	7590-VHVEG	Female	0	Yes	No	1	No	No phone service	DSL	No
1	5575-GNVDE	Male	0	No	No	34	Yes	No	DSL	Yes
2	3668-QPYBK	Male	0	No	No	2	Yes	No	DSL	Yes
3	7795-CFOCW	Male	0	No	No	45	No	No phone service	DSL	Yes
4	9237-HQITU	Female	0	No	No	2	Yes	No	Fiber optic	No
...	...	...	...	...	...	...	...	...	...	...
7038	6840-RESVB	Male	0	Yes	Yes	24	Yes	Yes	DSL	Yes
7039	2234-XADUH	Female	0	Yes	Yes	72	Yes	Yes	Fiber optic	No
7040	4801-JZAZL	Female	0	Yes	Yes	11	No	No phone service	DSL	Yes
7041	8361-LTMKD	Male	1	Yes	No	4	Yes	Yes	Fiber optic	No
7042	3186-AJIEK	Male	0	No	No	66	Yes	No	Fiber optic	Yes

7043 rows × 10 columns

Read ‘TelcomCustomer-Churn_2.csv’ as a DataFrame and assign it to a variable

df2=pd.read_csv("TelcomCustomer-Churn_2.csv")

df2

	customerID	OnlineBackup	DeviceProtection	TechSupport	StreamingTV	StreamingMovies	Contract	PaperlessBilling	PaymentMethod	MonthlyCharges	TotalCharges	Churn
0	7590-VHVEG	Yes	No	No	No	No	Month-to-month	Yes	Electronic check	29.85	29.85	No
1	5575-GNVDE	No	Yes	No	No	No	One year	No	Mailed check	56.95	1889.5	No
2	3668-QPYBK	Yes	No	No	No	No	Month-to-month	Yes	Mailed check	53.85	108.15	Yes
3	7795-CFOCW	No	Yes	Yes	No	No	One year	No	Bank transfer (automatic)	42.30	1840.75	No
4	9237-HQITU	No	No	No	No	No	Month-to-month	Yes	Electronic check	70.70	151.65	Yes
...	...	...	...	...	...	...	...	...	...	...	...	...
7038	6840-RESVB	No	Yes	Yes	Yes	Yes	One year	Yes	Mailed check	84.80	1990.5	No
7039	2234-XADUH	Yes	Yes	No	Yes	Yes	One year	Yes	Credit card (automatic)	103.20	7362.9	No
7040	4801-JZAZL	No	No	No	No	No	Month-to-month	Yes	Electronic check	29.60	346.45	No
7041	8361-LTMKD	No	No	No	No	No	Month-to-month	Yes	Mailed check	74.40	306.6	Yes
7042	3186-AJIEK	No	Yes	Yes	Yes	Yes	Two year	Yes	Bank transfer (automatic)	105.65	6844.5	No

7043 rows × 12 columns

Merge both the DataFrames on key ‘customerID’ to form a single DataFrame

df=pd.merge(df1, df2, on='customerID')

df

	customerID	gender	SeniorCitizen	Partner	Dependents	tenure	PhoneService	MultipleLines	InternetService	OnlineSecurity	...	DeviceProtection	TechSupport	StreamingTV	StreamingMovies	Contract	PaperlessBilling	PaymentMethod	MonthlyCharges	TotalCharges	Churn
0	7590-VHVEG	Female	0	Yes	No	1	No	No phone service	DSL	No	...	No	No	No	No	Month-to-month	Yes	Electronic check	29.85	29.85	No
1	5575-GNVDE	Male	0	No	No	34	Yes	No	DSL	Yes	...	Yes	No	No	No	One year	No	Mailed check	56.95	1889.5	No
2	3668-QPYBK	Male	0	No	No	2	Yes	No	DSL	Yes	...	No	No	No	No	Month-to-month	Yes	Mailed check	53.85	108.15	Yes
3	7795-CFOCW	Male	0	No	No	45	No	No phone service	DSL	Yes	...	Yes	Yes	No	No	One year	No	Bank transfer (automatic)	42.30	1840.75	No
4	9237-HQITU	Female	0	No	No	2	Yes	No	Fiber optic	No	...	No	No	No	No	Month-to-month	Yes	Electronic check	70.70	151.65	Yes
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
7038	6840-RESVB	Male	0	Yes	Yes	24	Yes	Yes	DSL	Yes	...	Yes	Yes	Yes	Yes	One year	Yes	Mailed check	84.80	1990.5	No
7039	2234-XADUH	Female	0	Yes	Yes	72	Yes	Yes	Fiber optic	No	...	Yes	No	Yes	Yes	One year	Yes	Credit card (automatic)	103.20	7362.9	No
7040	4801-JZAZL	Female	0	Yes	Yes	11	No	No phone service	DSL	Yes	...	No	No	No	No	Month-to-month	Yes	Electronic check	29.60	346.45	No
7041	8361-LTMKD	Male	1	Yes	No	4	Yes	Yes	Fiber optic	No	...	No	No	No	No	Month-to-month	Yes	Mailed check	74.40	306.6	Yes
7042	3186-AJIEK	Male	0	No	No	66	Yes	No	Fiber optic	Yes	...	Yes	Yes	Yes	Yes	Two year	Yes	Bank transfer (automatic)	105.65	6844.5	No

7043 rows × 21 columns

df.to_csv("churn.csv",index=False)

Verify if all the columns are incorporated in the merged DataFrame by using simple comparison Operator in Python.

df.columns

Index(['customerID', 'gender', 'SeniorCitizen', 'Partner', 'Dependents',
       'tenure', 'PhoneService', 'MultipleLines', 'InternetService',
       'OnlineSecurity', 'OnlineBackup', 'DeviceProtection', 'TechSupport',
       'StreamingTV', 'StreamingMovies', 'Contract', 'PaperlessBilling',
       'PaymentMethod', 'MonthlyCharges', 'TotalCharges', 'Churn'],
      dtype='object')

df.columns&df1.columns

Index(['customerID', 'gender', 'SeniorCitizen', 'Partner', 'Dependents',
       'tenure', 'PhoneService', 'MultipleLines', 'InternetService',
       'OnlineSecurity'],
      dtype='object')

we can see that columns form df1 are merged in df

df.columns&df2.columns

Index(['customerID', 'OnlineBackup', 'DeviceProtection', 'TechSupport',
       'StreamingTV', 'StreamingMovies', 'Contract', 'PaperlessBilling',
       'PaymentMethod', 'MonthlyCharges', 'TotalCharges', 'Churn'],
      dtype='object')

we can see that columns form df2 are merged in df

Impute missing/unexpected values in the DataFrame

df=df.replace(r'^\s*$', np.nan, regex=True)

df.isnull().sum()

customerID           0
gender               0
SeniorCitizen        0
Partner              0
Dependents           0
tenure               0
PhoneService         0
MultipleLines        0
InternetService      0
OnlineSecurity       0
OnlineBackup         0
DeviceProtection     0
TechSupport          0
StreamingTV          0
StreamingMovies      0
Contract             0
PaperlessBilling     0
PaymentMethod        0
MonthlyCharges       0
TotalCharges        11
Churn                0
dtype: int64

df= df.dropna()

df.isnull().sum()

customerID          0
gender              0
SeniorCitizen       0
Partner             0
Dependents          0
tenure              0
PhoneService        0
MultipleLines       0
InternetService     0
OnlineSecurity      0
OnlineBackup        0
DeviceProtection    0
TechSupport         0
StreamingTV         0
StreamingMovies     0
Contract            0
PaperlessBilling    0
PaymentMethod       0
MonthlyCharges      0
TotalCharges        0
Churn               0
dtype: int64

Null values are droped from dataframe

churndf=df.copy()

Make sure all the variables with continuous values are of ‘Float’ type

churndf.dtypes

customerID           object
gender               object
SeniorCitizen         int64
Partner              object
Dependents           object
tenure                int64
PhoneService         object
MultipleLines        object
InternetService      object
OnlineSecurity       object
OnlineBackup         object
DeviceProtection     object
TechSupport          object
StreamingTV          object
StreamingMovies      object
Contract             object
PaperlessBilling     object
PaymentMethod        object
MonthlyCharges      float64
TotalCharges         object
Churn                object
dtype: object

monthlycharges,totalcharges,tenure are continous values changing its datatype to flot.

churndf['tenure'] = churndf['tenure'].astype(float)
churndf['MonthlyCharges'] = churndf['MonthlyCharges'].astype(float)
churndf['TotalCharges'] = churndf['TotalCharges'].astype(float)

churndf.dtypes

customerID           object
gender               object
SeniorCitizen         int64
Partner              object
Dependents           object
tenure              float64
PhoneService         object
MultipleLines        object
InternetService      object
OnlineSecurity       object
OnlineBackup         object
DeviceProtection     object
TechSupport          object
StreamingTV          object
StreamingMovies      object
Contract             object
PaperlessBilling     object
PaymentMethod        object
MonthlyCharges      float64
TotalCharges        float64
Churn                object
dtype: object

Create a function that will accept a DataFrame as input and return pie-charts for all the appropriate Categorical features. Clearly show percentage distribution in the pie-chart.

import plotly.express as px
# if user did not set categorical_features 
def plotpie(df,categorical_features = None,dropna= False):
    categorical_features = churndf.select_dtypes(['object','category']).columns.to_list()
    print(categorical_features)
    threshold=17
    # loop through the list of categorical_features 
    for cat_feature in categorical_features: 
        num_unique = churndf[cat_feature].nunique(dropna = dropna)
        num_missing = churndf[cat_feature].isna().sum()
        # prints pie chart and info if unique values below threshold 
        if num_unique <= threshold:
            print('Pie Chart for: ', cat_feature)
            print('Number of Unique Values: ', num_unique)
            print('Number of Missing Values: ', num_missing)
            fig = px.pie(churndf[cat_feature].value_counts(dropna = dropna), values=cat_feature, 
                 names = churndf[cat_feature].value_counts(dropna = dropna).index,title = cat_feature,template='ggplot2')
            fig.show()
        else: 
            print('Pie Chart for ',cat_feature,' is unavailable due high number of Unique Values ')
            print('Number of Unique Values: ', num_unique)
            print('Number of Missing Values: ', num_missing)
            print('\n')

plotpie(churndf)

['customerID', 'gender', 'Partner', 'Dependents', 'PhoneService', 'MultipleLines', 'InternetService', 'OnlineSecurity', 'OnlineBackup', 'DeviceProtection', 'TechSupport', 'StreamingTV', 'StreamingMovies', 'Contract', 'PaperlessBilling', 'PaymentMethod', 'Churn']
Pie Chart for  customerID  is unavailable due high number of Unique Values 
Number of Unique Values:  7032
Number of Missing Values:  0


Pie Chart for:  gender
Number of Unique Values:  2
Number of Missing Values:  0

Pie Chart for:  Partner
Number of Unique Values:  2
Number of Missing Values:  0

Pie Chart for:  Dependents
Number of Unique Values:  2
Number of Missing Values:  0

Pie Chart for:  PhoneService
Number of Unique Values:  2
Number of Missing Values:  0

Pie Chart for:  MultipleLines
Number of Unique Values:  3
Number of Missing Values:  0

Pie Chart for:  InternetService
Number of Unique Values:  3
Number of Missing Values:  0

Pie Chart for:  OnlineSecurity
Number of Unique Values:  3
Number of Missing Values:  0

Pie Chart for:  OnlineBackup
Number of Unique Values:  3
Number of Missing Values:  0

Pie Chart for:  DeviceProtection
Number of Unique Values:  3
Number of Missing Values:  0

Pie Chart for:  TechSupport
Number of Unique Values:  3
Number of Missing Values:  0

Pie Chart for:  StreamingTV
Number of Unique Values:  3
Number of Missing Values:  0

Pie Chart for:  StreamingMovies
Number of Unique Values:  3
Number of Missing Values:  0

Pie Chart for:  Contract
Number of Unique Values:  3
Number of Missing Values:  0

Pie Chart for:  PaperlessBilling
Number of Unique Values:  2
Number of Missing Values:  0

Pie Chart for:  PaymentMethod
Number of Unique Values:  4
Number of Missing Values:  0

Pie Chart for:  Churn
Number of Unique Values:  2
Number of Missing Values:  0

Share insights for Q2.c

Genderwise data is seprated almost equally.

There are 30% dependents in data.

90.3% have phoneservices.

out of the them 42.2% have multiple lines.

44.4% people uses fiber optic internet services 34.4% people use DSL and 21.6% people do not have internet services.

out of the people who have internet services 28.7% have online security.

34.5% have online backup 34.4% have device protection 29% have techsupport.

38.4% have streaming tv services.

38.8% have streaming movies services.

out of three types of contract most of the people opt for month to month subcribtion.

60% people choose paperless billing.

electronic mail is most prefered payment method.

there is little biasedness in churning 73.4% do not churn.

Encode all the appropriate Categorical features with the best suitable approach

churndf1=churndf.copy()

churndf1

	customerID	gender	SeniorCitizen	Partner	Dependents	tenure	PhoneService	MultipleLines	InternetService	OnlineSecurity	...	DeviceProtection	TechSupport	StreamingTV	StreamingMovies	Contract	PaperlessBilling	PaymentMethod	MonthlyCharges	TotalCharges	Churn
0	7590-VHVEG	Female	0	Yes	No	1.0	No	No phone service	DSL	No	...	No	No	No	No	Month-to-month	Yes	Electronic check	29.85	29.85	No
1	5575-GNVDE	Male	0	No	No	34.0	Yes	No	DSL	Yes	...	Yes	No	No	No	One year	No	Mailed check	56.95	1889.50	No
2	3668-QPYBK	Male	0	No	No	2.0	Yes	No	DSL	Yes	...	No	No	No	No	Month-to-month	Yes	Mailed check	53.85	108.15	Yes
3	7795-CFOCW	Male	0	No	No	45.0	No	No phone service	DSL	Yes	...	Yes	Yes	No	No	One year	No	Bank transfer (automatic)	42.30	1840.75	No
4	9237-HQITU	Female	0	No	No	2.0	Yes	No	Fiber optic	No	...	No	No	No	No	Month-to-month	Yes	Electronic check	70.70	151.65	Yes
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
7038	6840-RESVB	Male	0	Yes	Yes	24.0	Yes	Yes	DSL	Yes	...	Yes	Yes	Yes	Yes	One year	Yes	Mailed check	84.80	1990.50	No
7039	2234-XADUH	Female	0	Yes	Yes	72.0	Yes	Yes	Fiber optic	No	...	Yes	No	Yes	Yes	One year	Yes	Credit card (automatic)	103.20	7362.90	No
7040	4801-JZAZL	Female	0	Yes	Yes	11.0	No	No phone service	DSL	Yes	...	No	No	No	No	Month-to-month	Yes	Electronic check	29.60	346.45	No
7041	8361-LTMKD	Male	1	Yes	No	4.0	Yes	Yes	Fiber optic	No	...	No	No	No	No	Month-to-month	Yes	Mailed check	74.40	306.60	Yes
7042	3186-AJIEK	Male	0	No	No	66.0	Yes	No	Fiber optic	Yes	...	Yes	Yes	Yes	Yes	Two year	Yes	Bank transfer (automatic)	105.65	6844.50	No

7032 rows × 21 columns

# Creating dummy variables for the remaining categorical variables and dropping the level with big names.

# Creating dummy variables for the variable 'MultipleLines'
ml = pd.get_dummies(churndf1['MultipleLines'], prefix='MultipleLines')
# Dropping MultipleLines_No phone service column
ml1 = ml.drop(['MultipleLines_No phone service'], 1)
#Adding the results to the master dataframe
churndf1 = pd.concat([churndf1,ml1], axis=1)

# Creating dummy variables for the variable 'OnlineSecurity'.
os = pd.get_dummies(churndf1['OnlineSecurity'], prefix='OnlineSecurity')
os1 = os.drop(['OnlineSecurity_No internet service'], 1)
# Adding the results to the master dataframe
churndf1 = pd.concat([churndf1,os1], axis=1)

# Creating dummy variables for the variable 'OnlineBackup'.
ob = pd.get_dummies(churndf1['OnlineBackup'], prefix='OnlineBackup')
ob1 = ob.drop(['OnlineBackup_No internet service'], 1)
# Adding the results to the master dataframe
churndf1 = pd.concat([churndf1,ob1], axis=1)

# Creating dummy variables for the variable 'DeviceProtection'. 
dp = pd.get_dummies(churndf1['DeviceProtection'], prefix='DeviceProtection')
dp1 = dp.drop(['DeviceProtection_No internet service'], 1)
# Adding the results to the master dataframe
churndf1 = pd.concat([churndf1,dp1], axis=1)

# Creating dummy variables for the variable 'TechSupport'. 
ts = pd.get_dummies(churndf1['TechSupport'], prefix='TechSupport')
ts1 = ts.drop(['TechSupport_No internet service'], 1)
# Adding the results to the master dataframe
churndf1 = pd.concat([churndf1,ts1], axis=1)

# Creating dummy variables for the variable 'StreamingTV'.
st =pd.get_dummies(churndf1['StreamingTV'], prefix='StreamingTV')
st1 = st.drop(['StreamingTV_No internet service'], 1)
# Adding the results to the master dataframe
churndf1 = pd.concat([churndf1,st1], axis=1)

# Creating dummy variables for the variable 'StreamingMovies'. 
sm = pd.get_dummies(churndf1['StreamingMovies'], prefix='StreamingMovies')
sm1 = sm.drop(['StreamingMovies_No internet service'], 1)
# Adding the results to the master dataframe
churndf1 = pd.concat([churndf1,sm1], axis=1)

churndf1

	customerID	gender	SeniorCitizen	Partner	Dependents	tenure	PhoneService	MultipleLines	InternetService	OnlineSecurity	...	OnlineBackup_No	OnlineBackup_Yes	DeviceProtection_No	DeviceProtection_Yes	TechSupport_No	TechSupport_Yes	StreamingTV_No	StreamingTV_Yes	StreamingMovies_No	StreamingMovies_Yes
0	7590-VHVEG	Female	0	Yes	No	1.0	No	No phone service	DSL	No	...	0	1	1	0	1	0	1	0	1	0
1	5575-GNVDE	Male	0	No	No	34.0	Yes	No	DSL	Yes	...	1	0	0	1	1	0	1	0	1	0
2	3668-QPYBK	Male	0	No	No	2.0	Yes	No	DSL	Yes	...	0	1	1	0	1	0	1	0	1	0
3	7795-CFOCW	Male	0	No	No	45.0	No	No phone service	DSL	Yes	...	1	0	0	1	0	1	1	0	1	0
4	9237-HQITU	Female	0	No	No	2.0	Yes	No	Fiber optic	No	...	1	0	1	0	1	0	1	0	1	0
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
7038	6840-RESVB	Male	0	Yes	Yes	24.0	Yes	Yes	DSL	Yes	...	1	0	0	1	0	1	0	1	0	1
7039	2234-XADUH	Female	0	Yes	Yes	72.0	Yes	Yes	Fiber optic	No	...	0	1	0	1	1	0	0	1	0	1
7040	4801-JZAZL	Female	0	Yes	Yes	11.0	No	No phone service	DSL	Yes	...	1	0	1	0	1	0	1	0	1	0
7041	8361-LTMKD	Male	1	Yes	No	4.0	Yes	Yes	Fiber optic	No	...	1	0	1	0	1	0	1	0	1	0
7042	3186-AJIEK	Male	0	No	No	66.0	Yes	No	Fiber optic	Yes	...	1	0	0	1	0	1	0	1	0	1

7032 rows × 35 columns

collist =['Contract', 'PaymentMethod', 'gender', 'InternetService','PhoneService', 'PaperlessBilling', 'Churn', 'Partner', 'Dependents']
LE = LabelEncoder()
for i in collist:
    churndf1[i] = LE.fit_transform(churndf1[i])

# We have created dummies for the below variables, so we can drop them
churndf1 = churndf1.drop(['MultipleLines','OnlineSecurity','OnlineBackup','DeviceProtection','TechSupport','StreamingTV','StreamingMovies'], 1)

churndf1

	customerID	gender	SeniorCitizen	Partner	Dependents	tenure	PhoneService	InternetService	Contract	PaperlessBilling	...	OnlineBackup_No	OnlineBackup_Yes	DeviceProtection_No	DeviceProtection_Yes	TechSupport_No	TechSupport_Yes	StreamingTV_No	StreamingTV_Yes	StreamingMovies_No	StreamingMovies_Yes
0	7590-VHVEG	0	0	1	0	1.0	0	0	0	1	...	0	1	1	0	1	0	1	0	1	0
1	5575-GNVDE	1	0	0	0	34.0	1	0	1	0	...	1	0	0	1	1	0	1	0	1	0
2	3668-QPYBK	1	0	0	0	2.0	1	0	0	1	...	0	1	1	0	1	0	1	0	1	0
3	7795-CFOCW	1	0	0	0	45.0	0	0	1	0	...	1	0	0	1	0	1	1	0	1	0
4	9237-HQITU	0	0	0	0	2.0	1	1	0	1	...	1	0	1	0	1	0	1	0	1	0
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
7038	6840-RESVB	1	0	1	1	24.0	1	0	1	1	...	1	0	0	1	0	1	0	1	0	1
7039	2234-XADUH	0	0	1	1	72.0	1	1	1	1	...	0	1	0	1	1	0	0	1	0	1
7040	4801-JZAZL	0	0	1	1	11.0	0	0	0	1	...	1	0	1	0	1	0	1	0	1	0
7041	8361-LTMKD	1	1	1	0	4.0	1	1	0	1	...	1	0	1	0	1	0	1	0	1	0
7042	3186-AJIEK	1	0	0	0	66.0	1	1	2	1	...	1	0	0	1	0	1	0	1	0	1

7032 rows × 28 columns

churndf1 =churndf1.drop(['MultipleLines_No','OnlineSecurity_No','OnlineBackup_No','DeviceProtection_No','TechSupport_No',
                       'StreamingTV_No','StreamingMovies_No'], 1)

churndf1.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 7032 entries, 0 to 7042
Data columns (total 21 columns):
 #   Column                Non-Null Count  Dtype  
---  ------                --------------  -----  
 0   customerID            7032 non-null   object 
 1   gender                7032 non-null   int32  
 2   SeniorCitizen         7032 non-null   int64  
 3   Partner               7032 non-null   int32  
 4   Dependents            7032 non-null   int32  
 5   tenure                7032 non-null   float64
 6   PhoneService          7032 non-null   int32  
 7   InternetService       7032 non-null   int32  
 8   Contract              7032 non-null   int32  
 9   PaperlessBilling      7032 non-null   int32  
 10  PaymentMethod         7032 non-null   int32  
 11  MonthlyCharges        7032 non-null   float64
 12  TotalCharges          7032 non-null   float64
 13  Churn                 7032 non-null   int32  
 14  MultipleLines_Yes     7032 non-null   uint8  
 15  OnlineSecurity_Yes    7032 non-null   uint8  
 16  OnlineBackup_Yes      7032 non-null   uint8  
 17  DeviceProtection_Yes  7032 non-null   uint8  
 18  TechSupport_Yes       7032 non-null   uint8  
 19  StreamingTV_Yes       7032 non-null   uint8  
 20  StreamingMovies_Yes   7032 non-null   uint8  
dtypes: float64(3), int32(9), int64(1), object(1), uint8(7)
memory usage: 624.9+ KB

Split the data into 80% train and 20% test.

X = churndf1.drop(['Churn','customerID'], axis=1)

X.head()

	gender	SeniorCitizen	Partner	Dependents	tenure	PhoneService	InternetService	Contract	PaperlessBilling	PaymentMethod	MonthlyCharges	TotalCharges	MultipleLines_Yes	OnlineSecurity_Yes	OnlineBackup_Yes	DeviceProtection_Yes	TechSupport_Yes	StreamingTV_Yes	StreamingMovies_Yes
0	0	0	1	0	1.0	0	0	0	1	2	29.85	29.85	0	0	1	0	0	0	0
1	1	0	0	0	34.0	1	0	1	0	3	56.95	1889.50	0	1	0	1	0	0	0
2	1	0	0	0	2.0	1	0	0	1	3	53.85	108.15	0	1	1	0	0	0	0
3	1	0	0	0	45.0	0	0	1	0	0	42.30	1840.75	0	1	0	1	1	0	0
4	0	0	0	0	2.0	1	1	0	1	2	70.70	151.65	0	0	0	0	0	0	0

y = churndf1['Churn'] #target variable

y.head()

0    0
1    0
2    1
3    0
4    1
Name: Churn, dtype: int32

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=100)

Normalize/Standardize the data with the best suitable approach.

scaler = StandardScaler()

X_train[['tenure','MonthlyCharges','TotalCharges']] = scaler.fit_transform(X_train[['tenure','MonthlyCharges','TotalCharges']])
X_test[['tenure','MonthlyCharges','TotalCharges']] = scaler.fit_transform(X_test[['tenure','MonthlyCharges','TotalCharges']])
X_train.head(5)

	gender	SeniorCitizen	Partner	Dependents	tenure	PhoneService	InternetService	Contract	PaperlessBilling	PaymentMethod	MonthlyCharges	TotalCharges	MultipleLines_Yes	OnlineSecurity_Yes	OnlineBackup_Yes	DeviceProtection_Yes	TechSupport_Yes	StreamingTV_Yes	StreamingMovies_Yes
1010	1	0	0	0	-1.201477	1	0	0	1	3	-0.483666	-0.937173	0	1	0	0	0	0	0
2732	0	0	0	1	-1.160653	1	2	0	0	1	-1.472880	-0.966413	0	0	0	0	0	0	0
2940	1	1	0	0	0.839728	0	0	0	1	2	-0.340687	0.249914	0	0	1	1	0	1	1
859	1	0	1	1	1.615386	0	0	2	1	1	0.003460	1.071889	0	1	1	1	1	1	1
2895	0	0	1	0	0.186543	1	2	1	1	0	-1.491168	-0.691863	0	0	0	0	0	0	0

Train a model using XGBoost. Also print best performing parameters along with train and test performance.

import xgboost as xgb
from xgboost import plot_tree

xg_reg = xgb.XGBClassifier()
xg_reg.fit(X_train,y_train)

preds = xg_reg.predict(X_test)
predictedlabels = xg_reg.predict(X_train)
print('Accuracy on Test data:',xg_reg.score(X_test, y_test) )
print('Accuracy on Training data:',xg_reg.score(X_train, y_train) )

Accuracy on Test data: 0.7704335465529495
Accuracy on Training data: 0.9454222222222223

accuracy on testing data is 77% and its 94% on traning data this is because its overfitting the data.

fig, ax = plt.subplots(figsize=(50,100))
xgb.plot_tree(xg_reg,rankdir='LR', ax=ax)
plt.show()

print("confusion matrix for test data")
cm = confusion_matrix(y_test, preds, labels=[0, 1])
df_cm = pd.DataFrame(cm, index = [i for i in ['no','yes']],
                  columns = [i for i in ['no','yes']])
plt.figure(figsize = (7,5))
sns.heatmap(df_cm, annot=True ,fmt='g')
plt.show()

confusion matrix for test data

print("classification  Matrix of testing data:\n",classification_report(y_test,preds))

classification  Matrix of testing data:
               precision    recall  f1-score   support

           0       0.83      0.86      0.84      1013
           1       0.60      0.54      0.57       394

    accuracy                           0.77      1407
   macro avg       0.71      0.70      0.71      1407
weighted avg       0.76      0.77      0.77      1407

print("confusion matrix for training data")
cm = confusion_matrix(y_train, predictedlabels, labels=[0, 1])
df_cm = pd.DataFrame(cm, index = [i for i in ['no','yes']],
                  columns = [i for i in ['no','yes']])
plt.figure(figsize = (7,5))
sns.heatmap(df_cm, annot=True ,fmt='g')
plt.show()

confusion matrix for training data

print("classification  Matrix of training data:\n",classification_report(y_train,predictedlabels))

classification  Matrix of training data:
               precision    recall  f1-score   support

           0       0.95      0.97      0.96      4150
           1       0.92      0.87      0.89      1475

    accuracy                           0.95      5625
   macro avg       0.94      0.92      0.93      5625
weighted avg       0.94      0.95      0.94      5625

print(pd.DataFrame(xg_reg.feature_importances_*100,columns =["IMP"],index = X_train.columns))

                            IMP
gender                 1.947350
SeniorCitizen          2.937410
Partner                1.653309
Dependents             2.643896
tenure                 3.762555
PhoneService           2.680086
InternetService       10.844049
Contract              44.031151
PaperlessBilling       3.093621
PaymentMethod          2.250037
MonthlyCharges         2.921533
TotalCharges           2.404255
MultipleLines_Yes      2.346547
OnlineSecurity_Yes     2.612715
OnlineBackup_Yes       2.511072
DeviceProtection_Yes   1.662004
TechSupport_Yes        4.154103
StreamingTV_Yes        2.370469
StreamingMovies_Yes    3.173835

contract is acting as root node in the tree. contract has more important then any other columns.

We can can perform hyperparameter tunning to get the optimal results.

Improve performance of the XGBoost as much as possible. Also print best performing parameters along with train and test performance.

X_train = X_train.drop(['gender','Partner','DeviceProtection_Yes'], axis=1)
X_test = X_test.drop(['gender','Partner','DeviceProtection_Yes'], axis=1)

removing columns with low importance

# A parameter grid for XGBoost
params = {'subsample': [0.6, 0.8, 1.0],
        'max_depth': range (2, 10, 1),
    'n_estimators': range(60, 220, 40),
    'learning_rate': [0.1, 0.01, 0.05]
        }

folds = 3
clf=xgb.XGBClassifier()
grid_search = GridSearchCV(clf, param_grid=params, scoring='roc_auc', cv=folds, verbose=1)

grid_search.fit(X_train,y_train)

Fitting 3 folds for each of 288 candidates, totalling 864 fits

GridSearchCV(cv=3,
             estimator=XGBClassifier(base_score=None, booster=None,
                                     callbacks=None, colsample_bylevel=None,
                                     colsample_bynode=None,
                                     colsample_bytree=None,
                                     early_stopping_rounds=None,
                                     enable_categorical=False, eval_metric=None,
                                     gamma=None, gpu_id=None, grow_policy=None,
                                     importance_type=None,
                                     interaction_constraints=None,
                                     learning_rate=None, max_bin=None,
                                     max_ca...
                                     max_leaves=None, min_child_weight=None,
                                     missing=nan, monotone_constraints=None,
                                     n_estimators=100, n_jobs=None,
                                     num_parallel_tree=None, predictor=None,
                                     random_state=None, reg_alpha=None,
                                     reg_lambda=None, ...),
             param_grid={'learning_rate': [0.1, 0.01, 0.05],
                         'max_depth': range(2, 10),
                         'n_estimators': range(60, 220, 40),
                         'subsample': [0.6, 0.8, 1.0]},
             scoring='roc_auc', verbose=1)

print('\n Best estimator:')
print(grid_search.best_estimator_)
print('\n Best hyperparameters:')
print(grid_search.best_params_)
print(grid_search.best_score_ )

 Best estimator:
XGBClassifier(base_score=0.5, booster='gbtree', callbacks=None,
              colsample_bylevel=1, colsample_bynode=1, colsample_bytree=1,
              early_stopping_rounds=None, enable_categorical=False,
              eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',
              importance_type=None, interaction_constraints='',
              learning_rate=0.05, max_bin=256, max_cat_to_onehot=4,
              max_delta_step=0, max_depth=3, max_leaves=0, min_child_weight=1,
              missing=nan, monotone_constraints='()', n_estimators=100,
              n_jobs=0, num_parallel_tree=1, predictor='auto', random_state=0,
              reg_alpha=0, reg_lambda=1, ...)

 Best hyperparameters:
{'learning_rate': 0.05, 'max_depth': 3, 'n_estimators': 100, 'subsample': 0.6}
0.8492384584455629

xgbs= xgb.XGBClassifier(base_score=0.5, booster='gbtree', callbacks=None,
              colsample_bylevel=1, colsample_bynode=1, colsample_bytree=1,
              early_stopping_rounds=None, enable_categorical=False,
              eval_metric=None, gamma=0, gpu_id=-1, grow_policy='depthwise',
              importance_type=None, interaction_constraints='',
              learning_rate=0.05, max_bin=256, max_cat_to_onehot=4,
              max_delta_step=0, max_depth=3, max_leaves=0, min_child_weight=1,
                monotone_constraints='()', n_estimators=100,
              n_jobs=0, num_parallel_tree=1, predictor='auto', random_state=0,
              reg_alpha=0, reg_lambda=1)
xgbs.fit(X_train,y_train)

preds = xgbs.predict(X_test)
predictedlabels = xgbs.predict(X_train)
print('Accuracy on Test data:',xgbs.score(X_test, y_test) )
print('Accuracy on Training data:',xgbs.score(X_train, y_train) )

Accuracy on Test data: 0.7882018479033405
Accuracy on Training data: 0.8163555555555555

fig, ax = plt.subplots(figsize=(50,100))
#xgb.plot_tree(xgbs, ax=ax)
xgb.plot_tree(xgbs,rankdir='LR',ax=ax)
plt.show()

print(pd.DataFrame(xgbs.feature_importances_*100,columns =["IMP"],index = X_train.columns))

                           IMP
SeniorCitizen         3.556906
Dependents            1.966765
tenure                8.688095
PhoneService          1.307482
InternetService       8.239846
Contract             33.494629
PaperlessBilling      7.514754
PaymentMethod         3.629251
MonthlyCharges        6.339473
TotalCharges          3.347968
MultipleLines_Yes     0.000000
OnlineSecurity_Yes    6.407682
OnlineBackup_Yes      2.072008
TechSupport_Yes       5.841285
StreamingTV_Yes       3.979757
StreamingMovies_Yes   3.614090

print("confusion matrix for test data")
cm = confusion_matrix(y_test, preds, labels=[0, 1])
df_cm = pd.DataFrame(cm, index = [i for i in ['no','yes']],
                  columns = [i for i in ['no','yes']])
plt.figure(figsize = (7,5))
sns.heatmap(df_cm, annot=True ,fmt='g')
plt.show()

confusion matrix for test data

print("classification  Matrix of testing data:\n",classification_report(y_test,preds))

classification  Matrix of testing data:
               precision    recall  f1-score   support

           0       0.83      0.89      0.86      1013
           1       0.65      0.53      0.58       394

    accuracy                           0.79      1407
   macro avg       0.74      0.71      0.72      1407
weighted avg       0.78      0.79      0.78      1407

print("confusion matrix for training data")
cm = confusion_matrix(y_train, predictedlabels, labels=[0, 1])
df_cm = pd.DataFrame(cm, index = [i for i in ['no','yes']],
                  columns = [i for i in ['no','yes']])
plt.figure(figsize = (7,5))
sns.heatmap(df_cm, annot=True ,fmt='g')
plt.show()

confusion matrix for training data

print("classification  Matrix of training data:\n",classification_report(y_train,predictedlabels))

classification  Matrix of training data:
               precision    recall  f1-score   support

           0       0.85      0.92      0.88      4150
           1       0.69      0.54      0.61      1475

    accuracy                           0.82      5625
   macro avg       0.77      0.73      0.74      5625
weighted avg       0.81      0.82      0.81      5625

after optimizing the model we can see that accuracy increased to 79% .

most important parameter is still contract.

training accuarcy is also droped suggesting data is not overftting.

Part B

1. Build a simple ML workflow which will accept a single ‘.csv’ file as input and return a trained base model that can be used for predictions. You can use 1 Dataset from Part 1 (single/merged).
2. Create separate functions for various purposes.
3. Various base models should be trained to select the best performing model.
4. Pickle file should be saved for the best performing model.

Include best coding practices in the code: • Modularization • Maintainability • Well commented code etc.

def dfinput(): 
    df=pd.read_csv(input("enter csv file name: ")) # user input for file name
    return df

def nonull(df): #dropping the null values
    df=df.replace(r'^\s*$', np.nan, regex=True)
    df= df.dropna()
    return df

def dttypes(df): #chaning the data type
    df['tenure'] = df['tenure'].astype(float)
    df['MonthlyCharges'] = df['MonthlyCharges'].astype(float)
    df['TotalCharges'] = df['TotalCharges'].astype(float)
    return df

def encodes(df): #labeling the columns
    # Creating dummy variables for the variable 'MultipleLines'
    ml = pd.get_dummies(df['MultipleLines'], prefix='MultipleLines')
    # Dropping MultipleLines_No phone service column
    ml1 = ml.drop(['MultipleLines_No phone service'], 1)
    #Adding the results to the master dataframe
    df = pd.concat([df,ml1], axis=1)
    # Creating dummy variables for the variable 'OnlineSecurity'.
    os = pd.get_dummies(df['OnlineSecurity'], prefix='OnlineSecurity')
    os1 = os.drop(['OnlineSecurity_No internet service'], 1)
    # Adding the results to the master dataframe
    df = pd.concat([df,os1], axis=1)
    # Creating dummy variables for the variable 'OnlineBackup'.
    ob = pd.get_dummies(df['OnlineBackup'], prefix='OnlineBackup')
    ob1 = ob.drop(['OnlineBackup_No internet service'], 1)
    # Adding the results to the master dataframe
    df = pd.concat([df,ob1], axis=1)

    # Creating dummy variables for the variable 'DeviceProtection'. 
    dp = pd.get_dummies(df['DeviceProtection'], prefix='DeviceProtection')
    dp1 = dp.drop(['DeviceProtection_No internet service'], 1)
    # Adding the results to the master dataframe
    df = pd.concat([df,dp1], axis=1)

    # Creating dummy variables for the variable 'TechSupport'. 
    ts = pd.get_dummies(df['TechSupport'], prefix='TechSupport')
    ts1 = ts.drop(['TechSupport_No internet service'], 1)
    # Adding the results to the master dataframe
    df = pd.concat([df,ts1], axis=1)

    # Creating dummy variables for the variable 'StreamingTV'.
    st =pd.get_dummies(df['StreamingTV'], prefix='StreamingTV')
    st1 = st.drop(['StreamingTV_No internet service'], 1)
    # Adding the results to the master dataframe
    df = pd.concat([df,st1], axis=1)

    # Creating dummy variables for the variable 'StreamingMovies'. 
    sm = pd.get_dummies(df['StreamingMovies'], prefix='StreamingMovies')
    sm1 = sm.drop(['StreamingMovies_No internet service'], 1)
    # Adding the results to the master dataframe
    df = pd.concat([df,sm1], axis=1)
    df = df.drop(['MultipleLines','OnlineSecurity','OnlineBackup','DeviceProtection','TechSupport','StreamingTV','StreamingMovies'], 1)
    collist =['Contract', 'PaymentMethod', 'gender', 'InternetService','PhoneService', 'PaperlessBilling', 'Churn', 'Partner', 'Dependents']
    LE = LabelEncoder()
    for i in collist:
        df[i] = LE.fit_transform(df[i])
    df =df.drop(['MultipleLines_No','OnlineSecurity_No','OnlineBackup_No','DeviceProtection_No','TechSupport_No',
                       'StreamingTV_No','StreamingMovies_No'], 1)
    return df

def split(df): #spliting the data
    X=df.drop(['Churn','customerID'], axis=1)
    y=df['Churn']
    
    return X,y

def scale(X,y):
    scaler = StandardScaler()
    X[['tenure','MonthlyCharges','TotalCharges']] = scaler.fit_transform(X[['tenure','MonthlyCharges','TotalCharges']])
    return X,y

def preprocess(df): #merging all the function into preprocessing function
    df=nonull(df)
    df=dttypes(df)
    df=encodes(df)
    X,y= split(df)
    X,y =scale(X,y)

    return X,y

def lr(X,y): # logistic regression function
    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=100)
    LR=LogisticRegression()
    LR.fit(X_train,y_train)
    lr.pllr = LR.predict(X_test)
    lr.srlr=LR.score(X_test, y_test)
    print("Accuracy of logistic regression",lr.srlr*100)
    return lr.srlr,lr.pllr

def knn(X,y): #KNN function
    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=100)
    KNN=KNeighborsClassifier(n_neighbors=5)
    KNN.fit(X_train,y_train)
    knn.plknn= KNN.predict(X_test)
    knn.srknn=KNN.score(X_test, y_test)
    print("Accuracy of KNN",knn.srknn*100)
    return knn.srknn,knn.plknn

def gn(X,y): # gaussian bayes function
    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=100)
    GN=GaussianNB()
    GN.fit(X_train,y_train)
    gn.plgn = GN.predict(X_test)
    gn.srgn=GN.score(X_test, y_test)
    print("Accuracy of GB",gn.srgn*100)
    return gn.srgn,gn.plgn
    

def svm(X,y): # SVM Function
    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=100)
    SVM = SVC(kernel='poly')
    SVM.fit(X_train,y_train)
    svm.plsvm = SVM.predict(X_test)
    svm.srsvm=SVM.score(X_test, y_test)
    print("Accuracy of SVM",svm.srsvm*100)
    return svm.srsvm,svm.plsvm
  

def dt(X,y): #decision tree function
    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=100)
    DT = DecisionTreeClassifier(criterion="entropy")
    DT.fit(X_train, y_train)
    dt.pldt = DT.predict(X_test)
    dt.srdt=DT.score(X_test, y_test)
    print("Accuracy of DT",dt.srdt*100)
    return dt.srdt,dt.pldt
  

def rst(X,y): # random forest function
    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=100)
    RST = RandomForestClassifier(criterion="entropy")
    RST.fit(X_train, y_train)
    rst.plrst = RST.predict(X_test)
    rst.srrst=RST.score(X_test, y_test)
    print("Accuracy of RST",rst.srrst*100)
    return rst.srrst,rst.plrst
  

def bestmodel(X,y): #best model finding function
    lr.srlr,lr.pllr = lr(X,y)
    knn.srknn,knn.plknn = knn(X,y)
    gn.srgn,gn.plgn = gn(X,y)
    svm.srsvm,svm.plsvm = svm(X,y)
    dt.srdt,dt.pldt = dt(X,y)
    rst.srrst,rst.plrst = rst(X,y)
    acc=[lr.srlr,knn.srknn,gn.srgn,svm.srsvm,dt.srdt,rst.srrst]
    name=['LR','KNN','GB','SVM','DT','RST']
    dic={}
    for i in range(len(name)):
        dic[name[i]] = acc[i]
    print(max(dic, key=dic.get),"is the best base model") #finding best model with accuracy
    filename = 'model'
    outfile = open(filename,'wb')# making pickle file for best model
    if max(dic, key=dic.get)=="LR":
        pickle.dump([lr.srlr,lr.pllr],outfile)
    elif max(dic, key=dic.get)=="KNN":
        pickle.dump([knn.srknn,knn.plknn],outfile)
    elif max(dic, key=dic.get)=="GB":
        pickle.dump([gn.srgn,gn.plgn],outfile)
    elif max(dic, key=dic.get)=="SVM":
        pickle.dump([svm.srsvm,svm.plsvm],outfile)
    elif max(dic, key=dic.get)=="DT":
        pickle.dump([dt.srdt,dt.pldt],outfile)
    else:
        pickle.dump([rst.srrst,rst.plrst],outfile)
    outfile.close()
    infile = open(filename,'rb')
    ndic=pickle.load(infile)
    print("loading data from pickle file for best model \n",ndic)
    
    

    
    
        
    

def main(): # merging all the functions
    df=dfinput()
    X,y = preprocess(df)
    bestmodel(X,y)
    

main() #calling main function

enter csv file name: churn.csv
Accuracy of logistic regression 78.25159914712152
Accuracy of KNN 74.84008528784648
Accuracy of GB 73.91613361762616
Accuracy of SVM 77.96730632551528
Accuracy of DT 72.21037668798863
Accuracy of RST 77.61194029850746
LR is the best base model
loading data from pickle file for best model 
 [0.7825159914712153, array([1, 0, 0, ..., 0, 0, 1])]

after calling the main function we gave input of merged data from part A question.

after running various models we found that logistic regression have the best accuarcy for the model.